Determining the regularization parameter: A hybrid reconstruction technique in fluorescence molecular tomography

Fluorescence molecular tomography (FMT) can obtain a sufficient data set and optimal three-dimensional images when projections are captured over 360°by CCD camera. In the Tikhonov regularization-based reconstruction procedure of FMT, the optimal regularization parameter obtained by some parameter-choice methods might miss the accurate result due to the model-mismatch and discretization error. Herein a two-step algorithm was proposed. In the first step, a suboptimal parameter was estimated based on the expected value of the Fourier coefficient of perturbation. Then in the second step, the L-curve criterion was adopted to get the optimal parameter in the permissible region obtained in the first step. With the optimal parameter and permissible region, more accurate reconstruction result was acquired. Experimental results suggested that such technique outperform the traditional L-curve criterion when applying into the FMT reconstruction.